EfficientNetV2-L (torchvision efficientnetv2l) fine-tuned for the Perturb subnet (netuid 26).
Open-weight model · Image classification
PlantDetect-V2-Preview
by Thenukegun10x thenukegun10x/PlantDetect-V2-Preview
PlantDetect-V2-Preview is an open-weight model for image classification from Thenukegun10x, released under other. It has 115M parameters. At 16-bit it needs about 0.3 GB of GPU memory, which fits on 1x MI300X from $1.85 an hour, at the lowest prices in the SAVRN Index. It draws 59 downloads a month.
Built with DINOv3. This is the completed V2 mean-pooling model for 2,991 plant species from Western Australia. Supply one photo, or 2–4 distinct photos from the same plant observation.
Parameters115M
Context—
Weights603.7 MB
Licenseother
AccessOpen weights
Monthly Downloads59
Runs On
What it takes to serve PlantDetect-V2-Preview (115M parameters): the memory its weights need at each precision, and the cheapest way to rent enough data-center GPUs to hold them.
| Precision | Weights | Memory needed | Cheapest setup | Per hour | Also fits |
|---|---|---|---|---|---|
| 16-bit | 0.2 GB | 0.3 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
| 8-bit | 0.1 GB | 0.1 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
| 4-bit | 0.1 GB | 0.1 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
Memory is the weights at that precision plus 20% for the runtime and a short context; a long context needs more. Prices are the lowest on-demand hourly rates in the SAVRN Index, read Oct 18, 2026.
PlantDetect-V2-Preview on every accelerator the SAVRN Index prices, at every precision
Model Card
Built with DINOv3. This is the completed V2 mean-pooling model for 2,991 plant species from Western Australia. Supply one photo, or 2–4 distinct photos from the same plant observation. This release uses the epoch-12 EMA weights from the expanded-data BF16 training run. The CLI uses self-contained TorchScript graphs (encoder.ts and classifier.ts) and imports no project modules. FP32 model.safetensors is also included for research use. Optimizer state and training data are excluded. executorch/PlantDetect-WA-V2-INT8.pte is a W8A8 export of this mean-pooling checkpoint for CPU inference without PyTorch. It uses custom plantquant:: kernels (INT8 weights with per-output-channel FP32 scales, FP16…
Excerpt from the card by Thenukegun10x, licensed other.
Identity and Version
- Repository
- thenukegun10x/PlantDetect-V2-Preview
- Publisher
- Thenukegun10x
- Task
- Image classification
- Modality
- Image
- Library
- pytorch
- Parameters
- 115M parameters
- Languages
- Not stated by the source
- Revision
- fd2bc109836dbdc5e2bb207f923c2595a0c40302
- First published
- 2026-09-25
- Last updated
- 2026-09-27
Files and Weights
55 files, 1.2 GB in total. The weights are 3 files totalling 603.7 MB in npz, safetensors.
Weights3 files · 603.7 MB
Configuration21 files · 395.5 KB
Documentation5 files · 52.6 KB
Other24 files · 631.9 MB
Repository2 files · 258 B
Every file
| File | Type | Size | SHA-256 |
|---|---|---|---|
| fp8/model.safetensors | Weights | 136.4 MB | 7f4ee321ebce |
| model.safetensors | Weights | 461.5 MB | d7de5dccf2e5 |
| species_centroids.npz | Weights | 5.7 MB | 79503c392093 |
| SHA256SUMS.json | Configuration | 5.4 KB | — |
| centroid_report.json | Configuration | 2.5 KB | — |
| config.json | Configuration | 111.3 KB | — |
| evaluation.json | Configuration | 1.8 KB | — |
| extension/moe/__init__.py | Configuration | 160 B | — |
| extension/moe/int8_ops.py | Configuration | 6.4 KB | — |
| extension/moe/mixed_encoder.py | Configuration | 3.1 KB | — |
| extension/moe/plant_ops.py | Configuration | 6.2 KB | — |
| extension/moe/test/test_plant_moe_ops.py | Configuration | 2.3 KB | — |
| fp8/adaround_report.json | Configuration | 29.2 KB | — |
| fp8/config.json | Configuration | 118.7 KB | — |
| plant_v2_cli.py | Configuration | 28.0 KB | — |
| src/__init__.py | Configuration | — | |
| src/models/__init__.py | Configuration | — | |
| src/models/expert_ffn.py | Configuration | 11.1 KB | — |
| src/models/moe_swiglu_v2.py | Configuration | 14.3 KB | — |
| src/models/plant_unified.py | Configuration | 6.6 KB | — |
| src/models/plant_vit_v2.py | Configuration | 16.3 KB | — |
| src/models/subcenter_arcface.py | Configuration | 4.1 KB | — |
| taxonomy_index.json | Configuration | 27.7 KB | — |
| torchscript_parity.json | Configuration | 479 B | — |
| NOTICE.md | Documentation | 217 B | — |
| README.md | Documentation | 13.0 KB | — |
| extension/moe/CPU_KERNEL_PLAN.md | Documentation | 25.4 KB | — |
| extension/moe/README.md | Documentation | 6.5 KB | — |
| licenses/DINOV3.md | Documentation | 7.5 KB | — |
| classifier.ts | Other | 20.2 MB | 6125829ddf39 |
| classifier_gpu.ts | Other | 20.2 MB | eae351952653 |
| encoder.ts | Other | 221.2 MB | 4366f22f547d |
| encoder_gpu.ts | Other | 221.2 MB | 02ed97e48b86 |
| executorch/PlantDetect-WA-V2-INT8.pte | Other | 136.5 MB | e28340a9a037 |
| executorch/linux-aarch64/plant_runner | Other | 4.6 MB | 0a710a11d73c |
| executorch/linux-x64/plant_runner | Other | 5.4 MB | 35daec582cbf |
| executorch/windows-x64/plant_runner.exe | Other | 2.5 MB | 52c3dd0c7198 |
| extension/moe/CMakeLists.txt | Other | 4.1 KB | — |
| extension/moe/build_linux.sh | Other | 1.8 KB | — |
| extension/moe/build_windows.ps1 | Other | 1.1 KB | — |
| extension/moe/compile.ps1 | Other | 13.9 KB | — |
| extension/moe/kernels/plant_fp16.cpp | Other | 22.8 KB | — |
| extension/moe/kernels/plant_int8.cpp | Other | 5.8 KB | — |
| extension/moe/kernels/plant_moe_cpu.cpp | Other | 1.8 KB | — |
| extension/moe/kernels/qgemm.h | Other | 18.5 KB | — |
| extension/moe/runner.cpp | Other | 8.5 KB | — |
| extension/moe/test/test_fp16_native.cpp | Other | 2.4 KB | — |
| extension/moe/test/test_moe_native.cpp | Other | 6.5 KB | — |
| licenses/APACHE-2.0.txt | Other | 11.4 KB | — |
| licenses/EXECUTORCH.txt | Other | 1.8 KB | — |
| licenses/PYTORCH.txt | Other | 3.5 KB | — |
| requirements-executorch-cli.txt | Other | 106 B | — |
| requirements.txt | Other | 146 B | — |
| .gitattributes | Repository | 237 B | — |
| .gitignore | Repository | 21 B | — |
License and Download
- License
- other
- Access
- Open weights, no gate
- Download size
- 603.7 MB
Download from Thenukegun10x
Released by Thenukegun10x through its official repository on Hugging Face. Read the license.
Memory Requirements
| Precision | Weights in memory |
|---|---|
| As published | 603.7 MB |
| 16-bit | 0.2 GB |
| 8-bit | 0.1 GB |
| 4-bit | 0.1 GB |
Weights only, from the published parameter count; the key-value cache and runtime add to this.
Questions About PlantDetect-V2-Preview
How much GPU memory does PlantDetect-V2-Preview need?
About 0.3 GB at 16-bit and 0.1 GB at 4-bit: the weights (115M parameters) plus a working margin. A long context needs more.
What is the cheapest GPU to run PlantDetect-V2-Preview on?
At 16-bit, 1x MI300X from $1.85 an hour; at 4-bit, 1x MI300X from $1.85 an hour, at the lowest on-demand prices the SAVRN Index lists.
What license is PlantDetect-V2-Preview released under?
other, as its publisher declares it. Read the license text before commercial use.
Similar Models
EfficientNetV2-L (torchvision efficientnetv2l) fine-tuned for the Perturb subnet (netuid 26).
EfficientNetV2-L (torchvision efficientnetv2l, 1000 ImageNet classes) fine-tuned with adversarial training for the Perturb subnet (netuid 26).
A BEiT-v2 image classification model. Trained on ImageNet-1k with self-supervised masked image modelling (MIM) using a VQ-KD encoder as a visual tokenizer (via OpenAI CLIP B/16 teacher). Fine-tuned on ImageNet-22k. - An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale: https://arxiv.org/abs/2010.11929v2 Explore the dataset and runtime metrics of this model in timm model results.
A Vision Transformer (ViT) image classification model. Trained on ImageNet-21k (with additional augmentation and regularization) in JAX by paper authors, ported to PyTorch by Ross Wightman. - How to train your ViT? Data, Augmentation, and Regularization in Vision Transformers: https://arxiv.org/abs/2106.10270 - An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale: https://arxiv.org/abs/2010.11929v2 Explore the dataset and runtime metrics of this model in timm model results.
SigLIP 2: Multilingual Vision-Language Encoders with Improved Semantic Understanding, Localization, and Dense Features https://arxiv.org/pdf/2502.14786 The model classifies each image into one of the following content categories: This model is intended for applications such as